Abstract
The series of La
.6
Sr
.4
Co
x
Fe
1-x
O
3
thin films were prepared by spray pyrolysis on fluorine-doped
tin oxide glass substrates and electrochemically characterized. LSCF 6455 was found to be the
best performing within the series, and showed improved oxygen reductive capabilities through
60 layers, which appeared to be a physical, rather than electrochemical limit. Staggered
annealing was shown to produce a more effective/mechanically stable catalyst.
Introduction
The reversible oxygen reduction and evolution reactions (ORR and OER)
4OH
-
O
2
+ 2H
2
O + 4e
-
(1)
are important reactions for many purposes,
1
including fuel cells
2
, metal-air batteries,
3
and in
photoelectrochemical cells.
4
However, both sides of the reaction proceed slowly, due to kinetic
stability of both water and gaseous oxygen.
5
The current catalysts used for these reactions are
finely ground platinum on a bed of carbon (Pt/C) for the leftward reaction (Oxygen Reduction
Reaction or ORR), and Iridium Oxide (IrO
2
) for the rightward reaction (Oxygen Evolution
Reaction or OER).
6
These catalysts are unsuitable for large scale industrial usage, as they
require precious metals, which are expensive and scarce.
6
For instance, the cost of Pt-based air
cathodes is responsible for the majority of the cost in zinc-air fuel cells.
3
As a result, there has
been a great deal of research interest focused on the creation of catalysts which can promote the
kinetics of the reactions without the need for expensive and scarce elements.
6
One promising class of catalysts includes metal oxides of the form ABO
3
, where A is a
lanthanide and B is a transition metal, both with a +3 charge.
1
While previous research has
shown the efficacy of synthesizing LaFe
x
Co
1-x
O
3
thin films through spray pyrolysis,
1
recent
research has shown that substitutions in both the A and B sites can produce increased catalytic
activity.
6,7
While other groups have tested individual mixtures of doubly substituted A and B
sites, no research has yet been done on a full LSCF series.
6
This may be due to the effort
involved in traditional synthesis of La-based metal oxides, which require exhaustive
complexation and sintering methods.
8,9
We found that within the La
.6
Sr
.4
Co
x
Fe
1-x
O
3
series, where
some of which had been previously investigated,
7
La
.6
Sr
.6
Co
.5
Fe
.5
O
3
was the best performing,
and showed increased performance through increased thickness up to a factor of ten higher than
LCFO electrodes.
1
The six compositions, 6491, 6482, 6464, 6455, 6446, and 6428, were chosen as they have
been previously characterized by X-ray diffraction, with all of the compositions except for 6455
most typically having a rhombohedral structure. 6491, 6482, and 6428 have also been observed
having a cubic crystal structure, while 6455 exists only as either an orthorhombic or tetragonal
crystal structure (International Centre for Diffraction Data, ICDD, powder diffraction file).
Results and Discussion
We prepared thin-film catalysts on fluorine-doped tin oxide (FTO) by spraying solutions of
metal nitrates with concentration 0.15M in both the A and B site in 10% ethanol and water.
Figure 1 shows the comparative OER and ORR graphs for each of the different materials.
Figure 1: Composite OER and ORR graphs of each of the six Co:Fe ratios tested at 12 layers
thick. All were run at 6 mV/s in oxygen-saturated 0.1 M NaOH solution for ORR and nitrogen-
saturated NaOH solution for OER vs. a saturated calomel reference electrode (SCE)..
We found that LSCF 6455 was the best performing in ORR, as it had the highest current density
per voltage applied through its peak. For OER, we found that 6455 was also the best performing.
While the ORR onset potential and peak potential appeared unpredictable by different ratios of
Co:Fe, the OER onset potentials appeared fairly consistent, with only 6446 having a different
onset. X-ray Diffraction was taken to ensure that we had prepared the proper crystal structure
and that the material was single phase. Figure 2 shows that the material is most likely LSCF
6455: however, the peaks are shifted similarly to the SnO
2
substrate. It is unclear what is
causing the shift and there is evidence that there is some Strontium Carbonate present (Figure 3).
We see in Figure 4 that increased thickness appears to show decreasing catalytic effect after 12
layers for OER, and decreasing catalytic effect between 12 and 20 layers, before improved
performance up to 40 layers thick for ORR. As we see in Figure 5B, however, at 40 layers the
electrode begins flaking off.
Figure 2
Figure 2: XRD of 12 layer 6455 film. Peaks are all shifted, and there appears a slight peak for
SrCO
3
Figure 3
Figure 3: Figure 2: LSCF 6455 (12 Layers) at 150x and 15000x magnifications BSE. Note the
Strontium rich areas, which appear as the dark spots on the 150x image. Below is an EDS
compilation of points on the surface which were more or less Sr rich with accompanying SEM
photo.
Figure 4:
Figure 4: Composite LSV OER and ORR graphs of different thicknesses of singly-annealed
LSCF 6455
Figure 5
Figure 5: LSCF 6455 at 100x Magnification 20 layers (A) and 40 layers (B) using an optical
microscope
To combat this flaking, staggered annealings were tested. We found that annealing after every 10
layers produced a film that was optically far more opaque (Figure 6) and functioned as a better
catalyst for both ORR and OER (Figure 7).
Figure 6
Figure 6: LSCF 6455 at 100x Magnification 2x10 layers and 4x10 layers
Figure 7
Figure 7: Composite LSV OER and ORR graphs of different thicknesses of multiply-annealed
LSCF 6455.
As compared to the Pt measurements in the prior paper, we see that the ORR of the new LSCF
6455, optimized for thickness, outperforms a Pt electrode by a factor of five for current density at
approximately the same peak.
1
For all LSV, the second scan is shown as the solutions were
cycled, because prior research
1
and initial experiments suggested that the first cycle of the first
ORR on a new electrode produced an unreproducible peak, while the second run was
reproducible after cycling through OER for all thicknesses save the thickest (70 layers in
staggered annealings of 10 at a time), which degraded to the point of unusability during the first
run.
Conclusions
La
.6
Sr
.4
Co
x
Fe
1-x
O
3
thin film metal oxides can be reliably produced through spray pyrolysis of
metal nitrates. Within the series, LSCF 6455 performed the best, and showed increased catalytic
activity as it was grown thicker up to 6x10 layers.
Experimental Methods
Oxide Electrode Preparation
Lanthanum(III) nitrate hexahydrate (>99.0% purity), strontium nitrate (>99% purity), iron(III)
nitrate nonahydrate (>98.5% purity), cobalt(II) nitrate hexahydrate (>98 % purity) were
purchased from Fluka, Aldrich, Sigma-Aldrich, and Alfa, respectively, and used without further
purification. Concentration was determined through serial dilution then Atomic Absorption
Spectroscopy and Inductively coupled plasma mass spectrometry through standard addition of
known standards. The believed 0.5 M solutions were instead found to be 0.4000 M (lanthanum),
0.5209 M (strontium), 0.2925 M (cobalt), and 0.3867 M (iron) (Figure 8).
Figure 8: Plot of Co Absorbance per mL 10 ppm standard added
Spray solutions of lanthanum/strontium/iron/cobalt nitrate were prepared by dissolving
the metal nitrates in water, then combining the aqueous nitrate solutions and diluting with doubly
deionized water and absolute EtOH to make solutions of 10% ethanol with La
3+
concentration of
0.9 M and Sr
2+
concentration of 0.6 M. The concentrations of Fe
3+
and Co
3+
ions were varied
(La:Sr:Fe:Co = 6:4:2:8, 6:4:4:6, 6:4:5:5, 6:4:6:4, 6:4:8:2, and 6:4:9:1). For example, to make 100
mL of the 6455 solution, 22.5 mL of lanthanum nitrate, 11.5 mL of strontium nitrate, 25.65 mL
of cobalt nitrate, and 19.4 mL of iron nitrate were dispensed with a buret together into a 100 mL
volumetric flask. 10 mL of ethanol was measured into a graduated cylinder and added, then
doubly deionized water was added to fill the volume.
An all glass sprayer built from a modified 10.00 mL glass pipette
1
was used to deposit the
solutions onto 3 inch square fluorine doped tin-oxide (FTO) sheets of glass (Hartford Glass). To
prepare the substrates for spraying, they were washed with Versa-Clean (Fisherbrand Versa-
Clean, 04-342), and rinsed with doubly deionized water. The sheets were then partially dried
with Kimwipes. After this, they were sprayed with 95% ethanol and wiped cleaned three times
on the fluorine doped side, and at least once on the non-fluorine doped side. After confirmation
of conductive side (measure of resistance >200 ohms), the conductive side of the substrate was
sprayed with 99% methanol and wiped with a lint-free cotton cloth.
After cleaning, the substrate was placed in a custom U-shaped bracket.
1
The conductive
side was sprayed after heating to between 250 and 280° C by allowing it to heat on a hot plate set
to 500 °C. Heating took between 25 and 40 minutes. It was noted that plates which had been
previously sprayed were more able to heat, reaching the upper end of the heating temperature
more quickly. The spray apparatus is otherwise used as described in reference 1.
A custom metal plate was placed on top of the glass substrate before spraying and
weighted with a chain and metal cylinder to prevent being blown (Figure 9). This ensured that an
even and controlled height of the plate was sprayed. Between each layer, the sprayer was turned
off for a count of 10 to allow the plate to reheat between sprays. The flow rate for the summer
ranged from 6.0 mL/s to 4.8 mL/s. After spraying, the hot plate was turned off to allow the plate
to cool. Once it was below 80 °C (typically closer to 40 or 50 °C), the plate was moved to an
oven (Ney 2-160 Series 2) at ambient temperatures. The oven contained a custom set of
aluminum shelves, which allow up to 6 samples to be heated at once. The oven was heated to
500 °C at a rate of 15 °C/min. The oven was set to remain at this temperature for 4 hours to
allow time for all of the nitrate solutions to burn off and for the formed metal-oxides to anneal.
The oven (which lacks a fan system) was allowed to cool itself, typically over a time span of 8
hours, and the samples were removed. To set the oven to cool itself, the second ramp rate is set
to 40 °C/min, the second temperature is set 100 °C, and the time for the second hold is 0.1 hours.
After this the oven turns itself off. The plates were then cut into strips approximately 1 cm thick.
The upper tip of the metal-oxide through the upper third of the newly cut and rinsed (with doubly
deionized water) electrodes were coated in clear fingernail polish (Rimmel 581 Clear) so the
FTO surface would not interfere with the electrochemical cell, while still allowing a conductive
connection to the apparatus.
Figure 9
Figure 9: Photo of the current spray setup with metal weight
Electrochemical Characterization
Once the nail polish had dried (felt dry to the touch, typically 5 minutes or less), the
electrodes were placed into a four-electrode electrochemical cell along with a Pt mesh counter
electrode and an SCE reference electrode. The cell contained 0.1 M NaOH electrolyte (replaced
weekly). Before each run, the solution was bubbled with oxygen gas (for ORR) or nitrogen gas
(for OER) for at least 10 minutes before data collection. Times longer than 10 minutes were not
seen to have a significant effect. The voltammetric experiments were performed at ambient room
temperature with a blanketing gas of the same as what had been bubbling in solution. The
electrochemical characterization for ORR was performed by sweeping from +0.1 V to between -
0.7 and -1.3 V (depending on material, trying to minimize the negative voltage while still seeing
a peak) at a rate of 6 mV/s. This sweep was controlled by a Pine Instruments bipotentiostat
(RDE5) and recorded with a PASCO Science Workshop (850 Interface) using the Pasco Data
Studio software. Current density as a function of voltage applied was calculated by dividing the
logged voltage by the geometric area of the electrode and multiplying by the current conversion
(.5). Two or three electrodes were produced and scanned for each composition and thickness to
test internal consistency within a plate. All electrodes were first run in ORR before moving and
running all electrodes for OER. When OER was run first, the electrochemical characterization
for OER was the same as above, except that the voltage swept between -0.2 V and +0.8 V. The
optimal La
.6
Sr
.4
Co
x
Fe
1-x
O
3
film composition was determined by creating comparative I-V curves
for both OER and ORR to see which composition created the highest current density at the
lowest voltage applied for ORR, and which created the highest current density at 0.8 V for OER.
All materials were run at both 8 and 12 layers, all showing marked ORR performance and
minimal OER performance in 12 layers.
Once LSCF 6455 had been found as the optimal ratio, its optimal thickness was found.
The various thicknesses were compared in the same way, producing comparative I-V curves.
The SCE was calibrated in the same 0.100 M NaOH electrolyte with the same
counter electrode and SCE. The hydrogen oxidation/reduction zero point was found in
both directions cycling between -0.96 V and -1.01 V at a rate of 10 mV/s, finding where the
current across a freshly polished (0.3 μm alumina) Pt flag electrode in blanketed/bubbled
hydrogen. The measured potential was −0.97015 V, so the potential versus RHE was
calculated by adding 0.97015 V to the potential measured versus SCE.
Materials Characterization
Materials in the paper were characterized identically to the paper this builds on, LaFe
x
Co
(1x)
O3
Thin-Film Oxygen Reduction Catalysts Prepared Using Spray Pyrolysis without Conductive
Additives (Dervishogullari, D.; Sharpe, C.,; Sharpe, L. ACS Omega 2017, 2, 11, 7695-7701), and
so the characterization is quoted from that paper.
“All measurements were done at Iowa State University’s Materials Analysis and Research
Laboratory. The SEM data were collected at 15 kV on an FEI Quanta-250 scanning electron
microscope with a field emission gun. All images are from backscattered electrons. The energy-
dispersive X-ray spectrum was acquired with an Oxford Aztec energy-dispersive spectrometer
system with an X-Max 80 detector having light element capability. Samples were examined at a
low-vacuum setting of 80 Pa of water except for the edge image due to charging of the glass
substrate. The XRD data were measured on a Siemens D 500 diffractometer with a Cu X-ray
tube operated at 45 kV and 30 mA with medium-resolution slits and a diffracted beam
monochromator.”
References
1: Dervishogullari, D.; Sharpe, C.,; Sharpe, L. ACS Omega 2017, 2, 11, 7695-7701
2: Gasteiger, H. A.; Baker, D. R.; Carter, R. N.; Wenbin, G.; Liu, Y.; Wagner, F. T.; Yu, P. T. In
Hydrogen Energy; Stolten, D., Ed.; Wiley-VCH: Weinheim, 2010;Chapter 1.
3: Neburchilov, V.; Haijiang, W.; Martin, J. J.; Qu, W. J. Power Sources 2010, 195, 1271 1291
DOI: 10.1016/j.jpowsour.2009.08.100
4: Gray, H. B. Nat. Chem. 2009, 1, 7 DOI: 10.1038/nchem.141
5: Jung, J.I.; Jeong, H.Y.; Lee, J.S.; Kim M.G.; Cho J.; Angew. Chem. Int. Ed. 2014, 53, 4582
4586
6: Zhu, H.; Zhang, P.; Dai, S. ACS Catal. 2015, 5, 6370−6385
7: Rincón, R.A.; Ventosa, E.; Tietz, F.; Masa, J.; Seisel, S.; Kuznetsov, V.; Schuhmann, W.;
ChemPhysChem 2014, 15, 2810 2816.
8: Sunarso, J.; Torriero, A. A. J.; Zhou, W.; Howlett, P. C.; Forsyth, M. J. Phys. Chem. C 2012,
116, 5827 5834 DOI: 10.1021/jp211946n
9: Wang, Y.; Cui, X.; Li, Y.; Chen, L.; Shu, Z.; Chen, H.; Shi, J. Dalton. Trans. 2013, 42, 9448
9452 DOI: 10.1039/c3dt50151e